Cellulose Hydrolysis Under Extremely Low Sulfuric Acid and ...infohouse.p2ric.org/ref/40/39049.pdf · 332 Kim, Lee, and Torget Applied Biochemistry and Biotechnology Vol. 91–93,

Effects of Acid and Temperature in Hydrolysis 331

Applied Biochemistry and Biotechnology Vol. 91–93, 2001

Copyright © 2001 by Humana Press Inc.All rights of any nature whatsoever reserved.0273-2289/01/91–93/0331/$12.50

331

*Author to whom all correspondence and reprint requests should be addressed.

Cellulose HydrolysisUnder Extremely Low Sulfuric Acidand High-Temperature Conditions

JUN SEOK KIM,1 Y. Y. LEE,*,1 AND ROBERT W. TORGET2

1Department of Chemical Engineering,230 Ross Hall, Auburn University, Auburn, AL 36849,

E-mail: [email protected];and 2National Renewable Energy Laboratory, Golden, CO, 80401

Abstract

The kinetics of cellulose hydrolysis under extremely low acid (ELA) con-ditions (0.07 wt%) and at temperatures >200°C was investigated using batchreactors and bed-shrinking flow-through (BSFT) reactors. The maximumyield of glucose obtained from batch reactor experiments was about 60% forα-cellulose, which occurred at 205 and 220°C. The maximum glucose yieldsfrom yellow poplar feedstocks were substantially lower, falling in the rangeof 26–50%. With yellow poplar feedstocks, a large amount of glucose wasunaccounted for at the latter phase of the batch reactions. It appears that asubstantial amount of released glucose condenses with nonglucosidic sub-stances in liquid. The rate of glucan hydrolysis under ELA was relativelyinsensitive to temperature in batch experiments for all three substrates. Thiscontradicts the traditional concept of cellulose hydrolysis and implies thatadditional factors influence the hydrolysis of glucan under ELA. In experi-ments using BSFT reactors, the glucose yields of 87.5, 90.3, and 90.8% wereobtained for yellow poplar feedstocks at 205, 220, and 235°C, respectively.The hydrolysis rate for glucan was about three times higher with the BSFTthan with the batch reactors. The difference of observed kinetics and perfor-mance data between the BSFT and the batch reactors was far above thatpredicted by the reactor theory.

Index Entries: Yellow poplar; cellulose hydrolysis; bed-shrinking flow-through reactor; kinetics.

Introduction

The acid-based treatment of biomass is gaining its position as a viablesaccharification process. Among the notable indicators of such is the

332 Kim, Lee, and Torget


emergence of the BCI biomass-to-ethanol plant (Jennings, LA) and the TotalHydrolysis Process of National Renewable Energy Laboratory (NREL),Golden, CO (1). The latter is built on three unique technical elements:employing extremely low acid (<0.1%) and high reaction temperature,applying a countercurrent moving bed scheme in the reactor design, andutilizing the bed-shrinking phenomena as a means to improve the reactorperformance.

There are distinct advantages of using extremely low acid (ELA) con-ditions for hydrolysis of lignocellulosic biomass. The low acidity mini-mizes the gypsum production, if any. The corrosion characteristics of ELAare close to those of neutral aqueous reaction so that standard-grade stain-less steel equipment can be used instead of high nickel alloy. ELA gives asignificant cost advantage in the equipment. A process using ELA alsoqualifies as a “green technology” because it has a minimal environmentaleffect. The recent advancement made in this technology has brought theacid hydrolysis process to a position where it can compete with the enzy-matic hydrolysis process in the overall process economics. This break-through technology has been proven on an ideally behaving bench-scalekinetic reactor system. An upscale experimental investigation using a con-tinuous countercurrent reactor is being conducted at NREL.

The ELA reaction conditions are beyond the region normally exploredin the conventional acid hydrolysis processes. Recent findings at NRELhave proven that yields in the vicinity of 90% are attainable under ELAconditions. The NREL data also suggest that the reaction mechanism maybe quite different in this region from those found in conventional processes.The present study was undertaken to provide further insights and kineticdata on the reactions taking place under ELA conditions.

Materials and Methods

Material

Yellow poplar sawdust feedstock was provided by NREL. The chemi-cal composition of a representative sample was 45.2% glucan, 15.8% xylan,and 18.4% Klason lignin. It was milled to pass through a 2-mm screen beforeuse. The composition of prehydrolyzed yellow poplar was 68.9% glucanand 26.9% Klason lignin. The prehydrolysis conditions were 174 and204°C/10 min using 0.07 wt% sulfuric acid by percolation reactor (2).

Batch Kinetic Experiments

All batch reactor experiments were performed using sealed tubularreactors. The reactors (13.5 cm3 of internal volume) were constructed out ofHastelloy C-276 tubing (0.5 in. [1.27 cm]). Both ends of the reactor werecapped with Swagelok end caps measuring 0.5 in. (1.27 cm) wide and 6 in.(15.24 cm) long. The reactors were packed with 0.8 g of solid substrate and8 mL of acid solution to achieve a solid-to-liquid ratio of 1:10. The sulfuricacid concentration was 0.07 wt% (pH 2.2). The reaction temperatures were



controlled in sand baths. The reactors were first submerged into a sand bathset at 50°C above the desired reaction temperature for rapid preheating.The reactors were then quickly transferred into another sand bath set at theprecise desired reaction temperature. The reactor temperature was moni-tored by a thermocouple inserted into the reactor. Reaction temperatures of205, 220, and 235°C were applied. After the desired reaction time, the reac-tion was quenched in an ice bath. The contents of the reactor were separatedinto liquid and solid by filtration and subjected to analyses.

Bed-Shrinking Flow-Through Kinetic Experiments

The bed-shrinking flow-through (BSFT) reactor system invented byNREL is described in Fig. 1. The main body of the reactor is Hastelloy C276tubing (2 in. [5.08 cm]). The internal volume was 294.4 cm3. The HastelloyC276 tubing (1/8 in. [1.6 mm] od × 0.03 in. [0.8 mm] id) was used to connectthe reactor with other components of the system as well as for the preheat-ing coil. The reactor, ancillary tubing, pump assembly, and collection sys-tem were connected and pressurized to 400 psig with N

2 gas. The flow rate

of the BSFT runs was kept at 30 mL/min. The amount of initial biomass was60 g. The reactor is equipped with an internal spring to compress the bedin the reactor as hydrolysis occurs. When the reaction reached the desiredtime, the flow was stopped and the reactor was quenched in cold water. The

Fig. 1. Laboratory setup for BSFT reactor system: 1, liquid tank; 2, metering pump;3, preheating coil; 4, bed-shrinking reactor; 5, thermometer; 6 and 7, temperature pro-grammable sand bath; 8, sampling port; 9, pressure holding tank; 10, N

2 gas; 11, acid

fluid inlet; 12, spring; 13, movable end; 14, compressed solid biomass; 15, liquid outlet.



liquid sample was collected from the liquid holding tank, and the remain-ing solid was taken from the reactor for further analysis of composition.

Analytical Methods

The sugars were determined by high-performance liquid chromatog-raphy using Bio-Rad Aminex, HPX-87P columns (3,4). A refractive indexdetector was used. The compositional analysis of all biomass solid sampleswas carried out by the NREL standard methods (5). The sugars in the liquidsample were determined after being subjected to a secondary acid hydroly-sis. The conditions of secondary hydrolysis were 4 wt% sulfuric acid, 121°C,and 1 h.

Results and Discussion

The ELA conditions have been applied mostly for hemicellulosehydrolysis, primarily as a method of pretreatment for the enzymatichydrolysis. For the past several years, however, it has been investigatedfrom a different angle and with a different purpose at NREL—as a meansof cellulose hydrolysis. This work has produced remarkable results in thatunusually high glucose yields have been achieved. Yields were particularlyhigh when the experiments were conducted with a BSFT reactor. Theobserved yields are far above the level projected by the known kinetics,often exceeding 90%. The high yields could not be explained solely from thereactor analysis. In our opinion, there are factors in the kinetics of hydroly-sis unique to ELA yet to be identified. Consequently, we became interestedin reconfirming the NREL experiments and in seeking explanations forthese findings.

Batch Reactions

The reaction kinetics under ELA is far from being established. Theliterature data on acid hydrolysis of glucan under the ELA conditions arecurrently limited to those of NREL. In the initial experiments, we con-ducted a series of batch runs for α-cellulose using 0.07 wt% of sulfuric acid,at varying temperatures of 205, 220, and 235°C. The reaction progress issummarized in Fig. 2. The results are shown in terms of the percentage ofglucan remaining in solid and the percentage of glucose released in liquid.The maximum yield of glucose obtained from the α-cellulose was about60% for 205 and 220°C. However, the maximum yield at 235°C was actually<40%. This is contrary to the conventional concept of cellulose hydrolysis,in which higher yields are obtained at higher temperature, because theactivation energy for hydrolysis is higher than that of the decompositionreaction.

The same batch experiments were also conducted using yellow poplaras the feedstock. The overall reaction profiles on the percentage of glucanremaining and the percentage of glucose released were similar to those ofα-cellulose runs (Fig. 3). The yield of glucose released increased slightly as



the temperature was raised. However, the maximum yield of glucose fromyellow poplar was much lower than that from α-cellulose for all three tem-peratures. The difference was most significant at 235°C, at which the maxi-mum observed batch yield was only 35.2% for yellow poplar feedstock, adrastic departure from the 59.2% observed for α-cellulose. Although therewas a similar tendency in the reaction profiles, the batch glucose yieldsobtained with prehydrolyzed (xylan-free) yellow poplar feedstocks were

Fig. 2. Semilog plot of cellulose remaining and glucose yield in batch reaction ofα-cellulose (0.07 wt% H

2SO

4).

Fig. 3. Semilog plot of glucan remaining and glucose yield in batch reaction ofuntreated yellow poplar (0.07 wt% H

2SO

4).



substantially lower than those of the untreated feedstocks for all tempera-tures (Fig. 4). This is not unique to the ELA conditions; it has been observedin hydrolysis with a higher acid level. The pretreated biomass had a higherfraction of crystalline cellulose, since the easily hydrolyzable glucan wasremoved during the prehydrolysis process. Prehydrolysis therefore makesthe feedstock more difficult to hydrolyze when it comes to acid hydrolysis.

The batch data clearly indicate that the glucose yields from lignocellu-losic feedstock (yellow poplar) are substantially lower than those fromα-cellulose. One of the potential reasons is that the extraneous materials inthe lignocellulosic biomass have a certain degree of buffering capacity foracids (6). The acidity of the liquid, and thus the reactivity, might have beenaffected. The data in Figs. 2–4, however, preclude this because the decaycurves of glucan are essentially the same for α-cellulose and yellow poplar.The hydrolysis reactivity is obviously not affected significantly by the pres-ence of the extraneous materials. Furthermore, the solid-to-liquid ratioapplied in the batch experiments was high enough (1:10) to mimic thebuffering effects of the biomass.

Another interesting point seen from the batch hydrolysis experimentsis that the glucan hydrolysis in solid was relatively insensitive to tempera-ture for all three substrates. The difference in overall slope, although notstraight lines, was <10% over the temperature span of 40°C. If one calcu-lates the activation energy for the glucan hydrolysis with the data obtainedin this study (ELA), it is approx 1 kcal/mol, an order of magnitude lowerthan those reported for conventional cellulose hydrolysis. This again con-tradicts the traditional cellulose hydrolysis and further indicates that thereare additional factors and reactions in the hydrolysis under ELA. The esti-mated rate of glucose decomposition in the traditional sense (glucose to

Fig. 4. Semilog plot of glucan remaining and glucose yield in batch reaction ofpretreated yellow poplar (0.07 wt% H

2SO

4).



hydroxymethyl furfural [HMF]) under the ELA conditions does not fullyaccount for the low yields observed with the yellow poplar feedstocks.What happens to the released glucose under the ELA conditions is uncer-tain. Some of it would decompose to HMF (7–9), some may recondensewith remaining cellulose or lignin (soluble and insoluble), and some mayrepolymerize as suggested by Conner et al. (10). To verify this point, weprepared a separate plot from the batch data. In this plot, the accountabilityof glucose, defined as glucan in solid plus the glucose in liquid, is plottedagainst time for both α-cellulose and yellow poplar (Fig. 5). The differencein the accountability is relatively small between the prehydrolyzed anduntreated yellow poplar (lower two curves). However, there is a substan-tial difference between α-cellulose and yellow poplar. The differenceof the accountability occurs only at the latter phase of the reaction, atwhich the soluble lignin and extraneous compounds tend to accumulateand the released glucose is at a high level. We also note that the true decom-position of glucose under acidic conditions is independent of the feedstock.These findings collectively indicate that the low yields for the lignocellu-losic feedstock are primarily owing to the interaction of the released glu-cose with the nonglucosidic compounds in the liquid. The unaccountedglucose would most likely exist in a condensed form with nonglucosidicsubstances, which includes the solubilized lignins. However, the existenceof the condensed products is yet to be proven. In any event, there is muchto be learned regarding the fundamental aspects of the reactions occurringunder the ELA conditions.

BSFT Reactor

A series of experiments were conducted using the BSFT reactorinvented by NREL. This reactor was used in its original design without any

Fig. 5. Semilog plot of glucose accountability for various substrates at 220°C.



Table 1Maximum Glucose Yields of Batch and BSFT Reactor

Maximum yield (%)/reaction time (min)

Reactor/feedstock 205°C 220°C 235°C

Batch (alpha-cellulose) 61.77/30 59.23/25 40.17/16Batch (pretreated yellow poplar) 26.62/16 35.45/13 20.43/10Batch (untreated yellow poplar) 49.82/16 50.98/16 35.22/13Bed-shrinking (untreated yellow poplar) 87.54/25 90.32/20 90.78/20

modification. The experiments were conducted at 205, 220, and 235°C using0.07 wt% H

2SO

4. The feedstock was untreated yellow poplar. The shrink-

ing-bed reactor operation was stopped at a desired point in the reactiontime; therefore, each run provided only one data point. For each reactioncondition, the experiments were repeated to obtain nine data points overthe reaction time. The results are summarized in Fig. 6 and Table 1.

The results we obtained are indeed astonishing in that the glucoseyields of 87.5, 90.3, and 90.8% were obtained at respective temperatures of205, 220, and 235°C. The concentrations of glucose in these runs were 2.25,2.37, and 2.47 wt%, respectively—certainly in a usable range. We haveessentially reproduced the results obtained at NREL involving BSFT reac-tors that can achieve yields in the vicinity of 90%. In addition, we expandedthe temperature range down to 205°C. Although the yield is somewhatlower at 87.5%, the low-temperature (205°C) run offers economic benefits

Fig. 6. Semilog plot of cellulose remaining and glucose yield for BSFT reactor,untreated yellow poplar (0.07 wt% H

2SO

4).



in other areas (lower equipment cost and energy input). It may prove to bea desirable operating condition.

We have previously conducted a modeling investigation that ascer-tains the positive effect of a bed-shrinking reactor (11). However, thedifference in observed kinetics and performance data between the BSFTand batch reactors is far above that predictable by the reactor theory inlieu of the solid-liquid contact pattern. One should also realize that thesimplistic approach of representing the acid hydrolysis of cellulose asone set of serial-parallel reaction patterns is grossly inadequate under theELA conditions.

The drastic difference in reactor performance between the batchand BSFT reactors is reaffirmed in Fig. 7. Here, the semilog plots for theremaining glucan are shown for the batch and BSFT reactors. It is clearlyseen that the hydrolysis rate for glucan (estimated from the initialslopes) is about three times higher with the BSFT reactor than with thebatch reactors. What causes this difference remains a mystery. For it to befully understood, the detailed reaction mechanism of this heterogeneouscatalytic reaction must be verified, and, therefore, further research isnecessary.

Acknowledgments

We wish to thank Nick Nagle of NREL for helpful guidance in thesetup and operation of bed-shrinking kinetic reactors. We also gratefullyacknowledge the financial support of the US Department of Energy(DE-FC36-99GO10475) and partial support by NREL (subcontract XGC-7-17041-01).

Fig. 7. Hydrolysis profiles of remaining glucan for batch and BSFT reactors at 235°C.



References

1. Torget, R., Hayward, T. K., and Elander, R. (1997), 19th Symposium on Biotechnologyfor Fuels and Chemicals, Colorado Springs, CO.

2. Torget, R., Hatzis, C., Hayward, T. K., Hsu, T., and Philippidis, G. P. (1996), Appl.Biochem. Biotechnol. 57/58, 85–101.

3. Grohmann, K., Himmel, M. E., Rivard, C., Tucker, M., Baker, J., and Torget, R. (1984),Biotechnol. Bioeng. Symp. 14, 137–157.

4. Yoon, H. H., Wu, Z., and Lee, Y. Y. (1995), Appl. Biochem. Biotechnol. 51/52, 5–19.5. Vinzant, T. B., Ponfick, L., Nagle, N., Ehrman, C. I., Reynolds, J. B., and Himmel,

M. E. (1994), Appl. Biochem. Biotechnol. 45/46, 611–626.6. Kim, S. B. and Lee, Y. Y. (1986), Biotechnol. Bioeng. Symp. 17, 71–84.7. Church, J. and Wooldridge, D. (1981), Ind. Eng. Chem. Prod. Res. Dev. 20, 371–378.8. Brenner, W. and Rugg, B. (1985), Report to Environmental Protection Agency, EPA/

600/S2-85/137, Washington, D.C.9. Bobleter, O., Schwald, W., Concin, R., and Binder, H. (1986), J. Carbohydr. Chem. 5(3),

387–399.10. Conner, A. H., Wood, B. F., and Hill, C. G. (1985), J. Wood Chem. Technol. 5, 461–489.11. Chen, R. W., Wu, Z. W., and Lee, Y. Y. (1998), Appl. Biochem. Biotechnol. 70/72, 37–49.

Documents

Cellulose Hydrolysis Under Extremely Low Sulfuric Acid and ...infohouse.p2ric.org/ref/40/39049.pdf · 332 Kim, Lee, and Torget Applied Biochemistry and Biotechnology Vol. 91–93,